OBJECTIVE To evaluate the evidence of analgesic efficacy of tramadol for the management of postoperative pain and the presence of associated adverse events in dogs. METHODS A comprehensive search using PubMed/MEDLINE, LILACS, Google Scholar and CAB databases with no restrictions on language and following a prespecified protocol was performed from June 2019 to July 2020. Included were randomized controlled trials (RCTs) performed in dogs that had undergone general anesthesia for any type of surgery. Two authors independently classified the studies, extracted data and assessed their risk of bias using Cochrane's tool. RevMan and GRADE methods were used to rate the certainty of evidence (CoE). CONCLUSIONS Overall 26 RCTs involving 848 dogs were included. Tramadol administration probably results in a lower need for rescue analgesia versus no treatment or placebo [moderate CoE; relative risk (RR): 0.47; 95% confidence interval (CI): 0.26-0.85; I2 = 0%], and may result in a lower need for rescue analgesia versus buprenorphine (low CoE; RR: 0.50; 95% CI: 0.20-1.24), codeine (low CoE; RR: 0.75; 95% CI: 0.16-3.41) and nalbuphine (low CoE; RR: 0.05; 95% CI: 0.00-0.72). However, tramadol administration may result in an increased requirement for rescue analgesia versus methadone (low CoE; RR: 3.45; 95% CI: 0.66-18.08; I2 = 43%) and COX inhibitors (low CoE; RR: 2.27; 95% CI: 0.68-7.60; I2 = 45%). Compared with multimodal therapy, tramadol administration may make minimal to no difference in the requirement for rescue analgesia (low CoE; RR: 1.12; 95% CI: 0.48-2.60; I2 = 0%). Adverse events were inconsistently reported and the CoE was very low. The overall CoE of the analgesic efficacy of tramadol for postoperative pain management in dogs was low or very low, and the main reasons for downgrading the evidence were risk of bias and imprecision.